Busca farmacofórica baseada em grupos químicos identifica moléculas promissoras contra as proteínas do SARS-CoV-2

Abstract

The SARS-CoV-2 pandemic highlighted the need for new therapeutic strategies, especially in the face of the emergence of variants with greater transmissibility and immune escape. In this context, virtual screening emerges as a promising in silico strategy for computer-aided drug discovery. Among the promising natural products in the development of new drugs, rutin stands out, studied for its pharmacological properties and antiviral activity. In this context, the present work aimed to perform a pharmacophore search, using chemical groups generated with rutin analogs as a model, aiming to identify molecules capable of inhibiting SARS- CoV-2 proteins. Initially, 220 molecules with structural similarity to rutin were downloaded from the PubChem database. For target selection, 4,124 three-dimensional structures of SARS-CoV-2 proteins were selected from the Protein Data Bank (PDB) and filtered by structural similarity, retaining those with RMSD >0.5 angstroms, resulting in 1,569 3D structures. From these structures, 193,561 pockets were predicted, with 4,000 retained having a druggability score ≥ 0.5. The 4,000 pockets and 220 selected ligands were prepared and subjected to docking, totaling 346,952 complexes generated by QuickVina 2, accounting for more than 3 million interactions with PLIP, generating 1,268 chemical groups, which were used for pharmacophore search. The pharmacophore search was performed using Pharmit in 10 databases, using the following filters: molecular weight range between 300 and 600, logP and number of rotatable bonds between 3 and 6, number of acceptor hydrogens from 4 to 12 and donor hydrogens from 0 to 6. This screening resulted in 174,359 molecules that were subjected to energy minimization and subsequent pharmacokinetic profile evaluation using ADMET-AI, observing that 67,912 molecules stood out with favorable ADMET profiles, presenting a better pharmacokinetic profile when compared to emergency drugs. Among the best molecules, 15 stood out for their superior affinity, ADMET, and toxicological safety parameters, interacting at the 8SDG-pocket 66, 8FEZ-pocket 228, 8H3E-pocket 2, 8H3E- pocket 3, and 8DRU-pocket 1 binding sites with the best ΔG (between -11 and -14.1 kcal/mol). Interaction characterization analysis demonstrated that the highlighted molecules exhibit amino acid residue interaction patterns similar to those described in the literature, suggesting a possible inhibitory effect against SARS-CoV-2. These findings highlight the relevance of these compounds as potential therapeutic candidates and are suggested for future in vivo and in vitro analyses. Furthermore, the results corroborate the potential of the hybrid virtual screening approach in the search for promising candidates for therapy against SARS- CoV-2.